eclipse convergence problem

[tinni]

i am working on a gas reservoir that has 40 years of production history. i have history matched the model. but when i am running my model for prediction it runs smoothly for next 20 years. but after that i am getting a problem message which is something like this
SOLUTION IN WELL HB-10 NOT CONVERGED AFTER 8 ITERATIONS
@ RESIDUAL ERROR IN OIL , WATER AND GAS FLOWS =
@ 5.40361E+04 -1.04940E-09 9.30135E-04
@ TRY INCREASING MXWSIT IN TUNING RECORD 3

what should i do now?...please help me

[06pg22]

try changing flow rates with realistic values attributed to well dia. Otherwise, smooth the SCAL curves.

[DAH7542]

Hey tinni, try this:

1. Tracking the source of the problem: include 'NEWTON=2' in RPTSCHED in the SCHEDULE section prior lunching the prediction run. This keywords allows to see a report of useful information in the PRT file to identify the problem. Use it and read the manual to understand what appears in the PRT file.
2. Pay attention when the run begins to slow down, i. e. when convergence problems appear. This can be done by using the previous keyword. You'll be able to identify the timestep at which convergence problems arise because it requires additional iterations, and finally, iteration limits stop the run as in your case.
3. The problem should not be addressed only by modifying the convergence limit criteria but by identifying the source of the problem, but this is your job. I tend to think that for your prediction run, convergence problems occur at a time-step when some changes are done to one or more wells (for example setting new well rates, or including new wells, etc). Check this. If this is the case you should reduce the maximum time step after a well change by modifyin item TMAXWC in TUNING keyword (Record 1). Try to change also the MXWSIT item in the same keyword as show by the simulator but once again, identify first the source of the convergence problem.

Good luck.